generative ai is a type of artificial intelligence that can generate new content including, text, images and code. it does this by learning patterns and structure of its training data to produce content with similar statistics.
in this concise and informative ’so’ video, we unravel the fascinating world of generative ai, explore its connection to machine learning and ponder its possibilities for the future.
if you have any questions you would like answered as part of our 'so' series, please get in touch and let us know.
our podcast, 'the digital marketing podcast' is now in video format on our youtube channel. make sure to subscribe for a brand new video podcast every week!
transcription:
so, what is generative ai?
we do like a new buzz term in digital marketing. let’s define ai first. artificial intelligence is not a single technology. there are loads of technologies that sit underneath artificial intelligence. we'll talk about machine learning in a moment. for example, generative ai is an ai that generates something, e.g. text.
chat gpt could be images, midjourney could be sound, it's just ai that generates something, and then we've come up with this new ‘generative ai’ term. there are other terms that we already had that would have kind of captured that anyway, really.
so, chatgpt is a generative ai?
yeah, it generates text, right? gpt itself is generative pre-trained transformer. basically, it takes stuff, fiddles it about and it outputs something else. it's essentially learned the sequence of words and it comes up with new ones. so it's a great example and that's when we started using this term. and then we had bard and all those different things as well come from google. they're all generative ais.
so, how is generative ai connected to machine learning?
so machine learning is generative ais. they take all the source of information out there and they basically understand how you're generating it. so you can kind of see that words come in this sequence.
so, it can see this is normally the kind of structure, it can untangle the patterns, and then it's able to repeat those afterwards as well. so machine learning is understanding the patterns. it seems smart. it's not smart, it just knows the patterns of things. it's cracked the recipe on how to anticipate what the next word needs to be in order to make perfect sense to an actual human.
that's the clever thing, but it is a bit of a pilot trick.
so, are we going to be seeing more of these generative ais?
yeah, it's worth understanding there's not an ai. there are loads of ais of which chat gpt is one, but there will be thousands. there'll be ones that are good with, images, sound, text, a number of different tasks and people will be finding ways of commercializing those.can they charge for a subscription? can they charge for the service base that they're using? can they charge for the output from that ai? there’ll be all sorts of different financial models, you'll have specialist ones and more generalist ones, but there's going to be lots and we'll be able to pick and choose.
we’ve started with chat gpt and things like that, but it's the start of a journey and we're going to see more and more of these things.